JPH0618240A - Pattern edge line estimating system and pattern inspection equipment - Google Patents

Abstract

PURPOSE:To obtain a system for estimating the edge line of pattern at high accuracy(and high speed) without increasing the number of pixels to be inspected, and a pattern inspection euipment employing the system. CONSTITUTION:When a binary encoded edge line is shifted one or more pixel from a decision criterion W1, a conductor pattern element just to the left of the decision criterion W1 is considered to be inadequate candidate pixel. Luminances (a), (b) of the inadequate candidate pixel and a pixel on the right thereof having no conductor pattern are taken out from a gray image and a position at which the inadequate candidate pixel traverses an edge line is then determined from an intersection (c) between a line connecting point (a), (b) and a threshold L.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an edge line estimation method suitable for inspecting a fine conductor pattern formed on a film carrier or the like, and a pattern inspection apparatus to which this method is applied.

[0002]

2. Description of the Related Art In a film carrier, which is widely used for mounting integrated circuits, a conductor pattern having a width of 0.1 to 0.5 mm is formed on a film such as polyimide. Figure 4
Shows a plan view of an example of a defect of the pattern. Conventionally, these patterns have been visually inspected with a microscope and require a great number of man-hours, so automation has been required.

A pattern matching method is considered as a method of automatically inspecting a pattern by a machine. The image of the pattern obtained by the TV camera is binarized, stored in the image memory, and compared with the standard pattern stored in advance.
This is a method of judging eligibility from the degree of coincidence of both patterns.

[0004]

FIG. 5 is a plan view showing a part of the pattern of the film carrier stored in the image memory. In the figure, each square represents one pixel, the hatched region is the conductor pattern, and the white background is the film only region. In an image obtained by irradiating light and photographing reflected light with a television camera, a conductor portion (hatching area) is bright, and a film without conductor or only a film is dark. When the image signal output from the television camera is binarized, the bright conductor pattern area "1" is coded as the dark area "0". The edge line of the binarized pattern is shown by a dot-and-dash line with reference numeral 13. Since it is binarized, it is "1" or "0" depending on the position where the edge line crosses.
To be judged. Therefore, the edge line is detected with an error of ± 0.5 pixel. This error is due to quantization. One pixel obtained by a television camera is 10μ
In the case of m, the edge line is detected with an accuracy of ± 5 μm, that is, 10 μm.

Now, assuming that the width of the conductor pattern on the film carrier is 50 μm, it is required to detect the chip 44 and the protrusion 45 in FIG. 4 with an accuracy of 3 μm.
However, the resolution obtained by an ordinary television camera is ± 5 μm, and the quantization error due to the binarization is ± 5 μm.
When combined with m, an error of ± 10 μm cannot be avoided.

When the pattern is enlarged and photographed in order to improve the detection accuracy of the edge line, the number of pixels for the inspection increases, and the time for the inspection becomes long and not practical. In FIG. 5, if it can be estimated how much the edge line 12 of the conductor pattern exceeds the determination standard W ₂ , the pattern can be inspected with high accuracy without increasing the number of inspection target pixels. There is no known method for estimating where the edge line crosses in.

Therefore, an object of the present invention is to provide a method for estimating an edge line of a pattern with high accuracy (at high speed) without increasing the number of pixels to be inspected, and a pattern inspection apparatus to which this method is applied.

[0008]

In order to solve the above problems, the present invention provides the following means.

A first means is a method of storing a grayscale image of a pattern in a pixel memory in advance and estimating a position where an edge line of the pattern crosses the pixel, and a pixel which the edge line crosses. From the brightness a of A, the brightness b of the pixel B adjacent to the pixel A having no pattern, and the threshold c (b <c <a), the pixel B from the center position of the pixel A
The edge line estimation method for a pattern is characterized in that the position α of the edge line deviated in the direction of the center position is calculated by the equation α = (ac) / (ab).

The second means stores the edge line of the non-defective standard pattern detected by the edge line estimation method in the standard pattern memory in advance and binarizes the grayscale image to generate a binary image. Value image memory, and the edge line of the standard pattern read from the standard pattern memory and the edge line of the binary image read from the binary image memory are compared by a pattern matching method, and the binary image memory A method of estimating an edge line of a pattern according to the first means, characterized in that a candidate for an unqualified pixel in the binary image read from the image is extracted, and the unqualified candidate pixel is set as a pixel A across which the edge line crosses. Is.

A third means is to photograph a conductor pattern or other fine pattern on the film carrier with a television camera to generate an image of the pattern,
The pattern of the image is binarized and stored in the binary memory as a binary pattern, and the edge line of the standard pattern is stored in advance in the standard pattern memory with the edge line of the allowable pattern as the criterion W _1. The amount of misalignment between the edge line of the binary pattern read from the binary memory and the edge line of the standard pattern read from the standard pattern memory is compared and determined by a determination criterion W ₁ to check the suitability of the pattern. A pattern inspection apparatus, comprising a grayscale image memory for storing the grayscale image of the image pattern obtained by the television camera,
The standard pattern memory stores in advance a judgment criterion W ₂ corresponding to an edge line in a narrower allowable range than the judgment criterion W ₁ , and determines a coordinate x of an unqualified pixel candidate on the edge line of the binary pattern. First, the luminance a of the unqualified candidate pixel in the grayscale image and the luminance b of the pixel in a region adjacent to the unqualified candidate pixel having no pattern are detected.
And the threshold value c (b <c <a) are used to calculate and estimate the coordinates of the edge line in the ineligible candidate pixel by the formula x + (ac) / (ab), and the edge line estimated by the calculation The pattern inspection apparatus is characterized by determining whether or not the pattern of the non-qualified candidate pixel is non-qualified by comparing the coordinates of 1) with the determination criterion W ₂ .

[0012]

In the pattern edge line estimation method according to the present invention, the position where the edge line crosses within one pixel is detected by using the brightnesses a and b and the threshold value c in the grayscale image. The threshold value c may be set adaptively by a well-known discriminant analysis method, or may be set to a predetermined fixed value. If the grayscale image is inspected for all the pixels on the edge line, the inspection time becomes long. Therefore, the binary image is inspected by the conventionally known pattern matching method to extract the unqualified candidate pixels, and the grayscale is applied only to the unqualified candidate pixels. Estimate the edge line based on the image. The pattern inspection apparatus of the present invention is an apparatus that inspects the suitability of a pattern by applying the edge line estimation method.

[0013]

EXAMPLES Next, the present invention will be described in more detail by way of examples. FIG. 1 is a diagram for explaining a pattern edge line estimation method according to an embodiment of the present invention. FIG. 1A is a diagram showing a part of an image obtained by photographing a conductor pattern with a television camera. In the figure, each square is one pixel and is indicated by reference numeral 10. The hatched area is a conductor pattern. A blank area without hatching is a portion without a conductor pattern, and there is only a film or nothing in this area (the film is punched out). x-3 to x + 4 indicate x coordinates, and y-4 to y
+5 indicates the y coordinate. W ₁ and W ₂ are criteria. Now, the edge line 12 of the conductor pattern is a curved line as shown in the figure, but when the edge line 12 is binarized, the edge line 13 becomes stepwise as shown by a chain line.

In this embodiment, the binarized edge line 13 and the judgment reference W ₁ are made to correspond by pattern matching, and pixels [x, y−1], [x, y], [x, y + 1],
At [x, y + 2], the binarized edge line 13 is set back from the criterion W ₁ . If tolerance is that it is one pixel, the edge line 12 to the inside of the four pixels, but is eligible if the left pixel from the edge line 12 is the determination reference _{W 2 [x-1, y} -1], [x -1, y], [x
If -1, y + 1] and [x-1, y + 2] are crossed, the conductor pattern is ineligible in this portion.

In this embodiment, it is possible to know whether the edge line 12 crosses the pixel on the left side of the criterion W ₂ and, if so, at which position. Since it takes too much time to perform such a detailed inspection on all of the edge lines, it is impractical to perform such detailed inspection. Therefore, there is a possibility that the allowable value may be exceeded by the pattern matching between the binarized edge line and the determination standard W _1. Is detected, and the pixel in that portion is selected as an unqualified candidate pixel. In the example of FIG. 1, pixels [x-1, y-1], [x-
1, y], [x-1, y + 1], and [x-1, y + 2] are ineligible candidate pixels.

Now, a grayscale image of a pattern in which the luminance is expressed by 4 bits and 16 gradations is stored in a grayscale image memory, and the luminance of the pixel whose y coordinate is y in the grayscale image (see FIG. 1).
FIG. 1B shows the luminance of the pixels on the line EE in FIG. a is an ineligible candidate pixel [y, x-
1], and b is the brightness of the pixel [x, y] that is adjacent to it and has no conductor pattern. In c, the threshold L is the luminance curve u
Represents a point that intersects with ₁ .

In the present embodiment, the threshold value L is set to a fixed value of 0.7 which is preset between the brightest brightness 1 and the darkest brightness. FIG. 1C shows a point a in FIG.
It is shown that the intersection point of the straight line connecting the line b and the threshold value L is obtained as c. If the x coordinate of the center of the pixel [x-1, y] is x and the x coordinate of the center of the pixel [x, y] is x + 1.0 (the length of one pixel is 1.0), the point c X coordinate x
Α in + α is α = (ac) / (ab) (1).

The x coordinate of this point c is the ineligible candidate pixel [x-
It has been found empirically that it corresponds to the x coordinate of the edge line 12 in [1, y]. The threshold L is usually 0.5 to
It is selected to be about 0.7. Further, the threshold L may be adaptively obtained by automatic binarization by a well-known discriminant analysis method.

In this way, the ineligible candidate pixel [x-1,
y] is crossed by the edge line 12 and its position is deeper than the criterion W _2. Therefore, this conductor pattern is defined by the pixel [x-1,
y], it was presumed to be ineligible. Similarly, the suitability can be estimated for other unqualified candidate pixels.

FIG. 2 is a block diagram showing the structure of the pattern inspection apparatus according to the present invention. This embodiment is an apparatus for precisely inspecting the edge line of a conductor pattern by applying the pattern edge line estimation method described with reference to FIG. 1, and includes a camera 1, an A / D conversion circuit 2, and a grayscale image memory 3. ,
A binarization circuit 4, a binary image memory 5, an edge detection circuit 6,
It is composed of a calculator 7 and a standard pattern memory 8.

The camera 1 is a television camera having a CCD image pickup device and picks up an image of a film carrier.
A conductor pattern having a width of 0.1 mm is formed on the film carrier. The output of the camera 1 is an analog grayscale image signal. The A / D conversion circuit 2 uses the analog image signal 101
Is converted into a 16-gradation digital image signal 102. The grayscale image memory 3 stores a digital image for one screen in 16 gradations. The binarization circuit 4 receives the 16-gradation digital image signal 103a read from the grayscale image memory 3, and generates a binary image signal 104. The binary image memory 5 stores the binary image signal 104 for one screen. The edge detection circuit 6 receives the binary image signal 105 read from the binary image memory 5, and generates a boundary line (edge line) of the conductor pattern represented by the binary image signal 105. The edge pattern signal 106 is a signal representing the x and y coordinates of the edge line. The standard pattern memory 8 stores in advance the standard of the conductor pattern of the film carrier to be inspected. The standard pattern is 2
There are types, the first of which represents the judgment criterion W ₁ of FIG. 1 and the _{second of} which represents the judgment criterion W ₂ of FIG. All standard patterns are stored in binary and represent edge lines.

The calculation unit 7 is a CPU. In the arithmetic unit 7, pattern matching is performed between the edge line of the pattern to be inspected represented by the edge line signal 106 and the first standard edge line read out from the standard pattern memory 8 and represented by the judgment reference W ₁ , and both edge lines are detected. The edge line of the pattern to be inspected is judged criteria W
When away ₁ from 1 pixel or more, (the pixel represented by "1" in binary) pixel of the last conductor pattern from W ₁ in one of the x-axis or y-axis away one pixel or more Is selected as an unqualified candidate pixel. In the example of the conductor pattern shown by hatching in FIG. 1A, as described with reference to FIG.
-1, y-1], [x-1, y], [x-1, y +
1] and [x-1, y + 2] are extracted as unqualified candidate pixels. When the non-qualified candidate pixel is extracted, the calculation unit 7 reads the gray-scale image of the non-qualified candidate pixel from the gray-scale image memory 3 in order to check how the edge line 12 crosses in the area. Luminance a and the luminance b of the pixel on the side of the criterion W ₁ adjacent to the non-qualified candidate pixel are extracted, and the coordinates of the edge line of the pixel are determined by the equation (1). And FIG.
In the example of (A), the obtained coordinates are on the left side of the second standard edge line represented by the determination criterion W ₂ , and the unqualified candidate pixels are determined to be unqualified.

FIG. 3 is a diagram showing the flow of processing performed by the arithmetic unit 7 in the embodiment of FIG. 2 described above.

In the embodiment of the pattern inspection apparatus described with reference to FIG. 2, the rough inspection performed by the binary image targets all edge lines, but the detailed inspection performed by the grayscale image is the inspection of the binary image. Since the inspection is performed only for the unqualified candidate pixels extracted in step 1, the entire inspection time is short in the short time close to that of the conventional apparatus that inspects only the binary image.

[0025]

As described above in detail with reference to the embodiments, according to the present invention, an edge line estimating method for a pattern capable of accurately estimating an edge line at high speed and an edge line estimating method for the edge line estimating method are used. It is possible to provide a pattern inspection apparatus capable of accurately and rapidly determining whether a pattern is appropriate or not by applying it to various estimations.

[Brief description of drawings]

FIG. 1 is a diagram illustrating an embodiment of a pattern edge line estimation method according to the present invention.

FIG. 2 is a diagram showing a configuration of an embodiment of a pattern inspection apparatus according to the present invention.

FIG. 3 is a diagram showing a flow of processing performed by a calculation unit 7 in FIG.

FIG. 4 is a plan view showing an example of a conductor pattern formed on a film carrier.

FIG. 5 is a view showing an example of a conductor pattern obtained by imaging a film carrier with a television camera.

[Explanation of Codes] 1 television camera 2 A / D exchange circuit 3 grayscale image memory 4 binarization circuit 5 binary image memory 6 edge detection circuit 7 arithmetic unit 8 standard pattern memory 10 pixels 12 conductor pattern edge line 13 2 Quantized edge lines 41 to 43 Conductor pattern 44 Missing 45 Protrusion

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁵ Identification number Internal reference number FI Technical display location G06F 15/70 335 Z 9071-5L H05K 3/00 Q 6921-4E

Claims (3)

[Claims] 1. A method for estimating a position where an edge line of the pattern crosses the pixel by storing a grayscale image of the pattern in a pixel memory in advance, the luminance a of a pixel A crossing the edge line. And the brightness b of the pixel B that is adjacent to the pixel A and has no pattern and the threshold value c (b <c <
From a), the position α of the edge line deviated from the center position of the pixel A in the direction of the center position of the pixel B is calculated by the equation α = (ac) / (ab). Edge line estimation method for patterns. 2. The edge line of the standard pattern created according to claim 1 is stored in advance in a standard pattern memory, the grayscale image is binarized to generate a binary image, and the binary image is stored in the binary image memory. The position of the edge line of the standard pattern read from the standard pattern memory and the position of the edge line of the standard pattern read from the binary image memory are compared, and the unqualified pixel in the binary image read from the binary image memory is compared. 2. The pattern edge line estimation method according to claim 1, wherein the candidate is extracted and the non-qualified candidate pixel is set as a pixel A across which the edge line crosses. 3. A conductor pattern or other fine pattern on a film carrier is photographed by a television camera to generate an image of the pattern, and the image pattern is binarized and stored in a binary memory as a binary pattern. Then, the edge line of the allowable pattern is determined as the criterion W ₁
As an example, an edge line of the standard pattern is stored in advance in the standard pattern memory, and the positional deviation amount between the edge line of the binary pattern read from the binary memory and the edge line of the standard pattern read from the standard pattern memory is stored. In a pattern inspection apparatus for inspecting the suitability of the pattern by a determination criterion W _{1 including} a grayscale image memory for storing the image pattern obtained by the television camera as a grayscale image, and the standard pattern memory Stores in advance a judgment criterion W ₂ corresponding to an edge line in a narrower allowable range than the judgment criterion W _1, and first detects the coordinate x of a candidate pixel that is not suitable for the edge line of the binary pattern, The brightness a of the unqualified candidate pixel in the grayscale image and the pattern adjacent to the unqualified candidate pixel are Luminance b and the threshold c of pixels have regions (b <c
<A) and the coordinates of the edge line in the ineligible candidate pixel are calculated and estimated by the formula x + (ac) / (ab), and the coordinates of the edge line estimated by the calculation are used as the determination criterion W. _2. A pattern inspection apparatus, which judges whether or not the pattern of the unqualified candidate pixel is unqualified by comparing with ₂ .